Hi All, I am setting Exchange 2010 DAG to span accross 2 Data Centres over Dark Fibre. As this is SAN I will only be hosting 2 copies of each database (i.e. one in each DC) I have a query regarding AutoDatabaseMountDial The default is BestAvailability which means it can lose upto 12 logs My question is, is there any reason why this should not be set to Lossless as ideally I don't want to lose any data? Also does using using BestAvailability mean it will always lose data or will it mount the db and still replay any logs that are on the other DAG members after to minimise data loss? I have also seen that other vendors like EMC bost they have techniques to have zero data loss via Synchronous SAN replication - any views on that? Thanks ECL

Lossless would mean if there are any missing logs it will not fail over. If not losing data is a requirement this is the correct setting. If potentially losing 'a little data' is acceptable then BestAvaialbility would allow for automatic failover isf between 0 and 12 log files were missing. The server will only replay any pending logs it has as the failover occurs, it's not going to look for additional logs from the old source once it has failved over. BestAvailability means it *could* lose data if log files are missing. EMC does SRDF replication I believe which is byte level replication rather than log shipping. Advantage... theoretically 0 data loss. Disadvantage byte level replication dutifully replicates database corruption so you can instantly failover with 0 data loss to a corrupt database which is equally fubar.

Thanks for your reply Chris, No one wants to lose data so this should be to "Lossless" ..... is this what you have it set to? and does everyone else too? If so I don't understand why microsoft don't have this set as the default unless they want the system to failover automatically regardless at the expense of data loss? i.e less support calls for them with users asking why its not failed over???? and accept the data loss... Thanks ECL

I work for a company that addresses the email continuity/recovery differently so how I have it set probably isn't applicable. :) Some (many?) organizations are willing to have some type of minimal data loss in exchange for availability. Since replication will often lag at times for a typical organization (especially if they are replicating over the WAN), automatic failover would have been hit or miss. By selecting the current default I assume the thinking was, failover will happen in more cases with minimal data loss. I guess that all depends on ones perspective. It's a risk tolerence question that organizations need to address individually as I don't think there's a universal right answer.

Just one more thing do you know the procedure or command to run to force the database to mount after a lossy failover? e.g. Autodatabasemountdial is set to Lossless. Server A containing the active copy suddenly crashed. Now Server B containing passive copy won't automatically mount as missing a log or 2. Then what command do I run to force it to mount? (obviously I will try and recover the Server A first to get try and get back all the logs but just incase I can't and it needs a rebuild so I will need to accept the loss and force Server B Database online) Thanks ECL